OpenAI Fixes ChatGPT Agent Flaw That Could Let Attackers Forge an AI Insider

OpenAI patches a ChatGPT agent vulnerability that could have allowed attackers to impersonate a trusted AI insider through prompt injection.